When we think of the air we breathe, we usually don’t think fluids. But air is a fluid. And bacteria and viruses are carried by fluids. So understanding the dynamics of fluids — how they flow under the influence of various forces, such as gravity and any initial momentum imparted to the fluid — is crucial to understanding how viruses or other pathogens spread from place to place, from person to person.

Lydia Bourouiba made the connection while studying fluid dynamics at McGill University in Montreal. In 2003, she was partway through her PhD when the SARS epidemic struck. She realized then that she wanted her work to have an impact on public health.

Bourouiba now leads the Fluid Dynamics of Disease Transmission Laboratory at the Massachusetts Institute of Technology (MIT). For more than a decade, she has focused her attention on how fluids can help disease move from one host or reservoir to the next. Armed with high-speed cameras, some fancy mathematics and old-fashioned grit, Bourouiba studies everything from the motion of droplets that are ejected when we breathe, cough or sneeze to how splashes of water droplets from leaves can spread pathogens from plant to plant. She explored the current state of such knowledge in papers in the 2021 Annual Review of Fluid Mechanics and the Annual Review of Biomedical Engineering.

Knowable Magazine spoke to Bourouiba on how understanding the dynamics of fluids can inform public health measures and help limit the spread of infectious diseases such as Covid-19.

This conversation has been edited for length and clarity.

How did you get interested in fluid dynamics and infectious diseases?

When I took my first class on fluid dynamics, I fell in love with the topic, because of its beauty and universality. The beauty is tied to the mathematics of fluid dynamics, and that you may find fluid processes working at the scale of stars and galaxies, as well as at the cellular level.

Despite the age of the field, so many fundamental questions remain to be answered. I find such universality and depth beautiful. I have also always felt strongly about human rights, and in particular about equity in access to public health and education. It’s another side of who I am.

A still created from a high-speed video camera and special lighting show the steep climb and fall of larger water droplets flushed from a toilet as well as the fainter trajectories of smaller droplets.

A toilet sends both large and small droplets of water aloft during a flush, as imaged here with the help of a high-speed video camera attached to the side of a toilet basin. The weight of larger droplets leads most to fall back down into the toilet or onto surfaces. Smaller droplets, which are more likely to form after the use of some cleaning products, can be cast up into the air, where they may remain suspended. Lydia Bourouiba’s studies suggest airborne droplets from toilets in hospitals could contribute to the spread of bacterial spores that cause a potentially deadly form of diarrhea.


Coming back to fluids: Pathogens travel in fluids, whether they’re in the body or outside, and through air, which is a fluid as well. In combining fluid dynamics with applied questions about disease transmission and other topics, I saw a way to apply myself to fundamental open questions in fluid dynamics and mathematics. After much exploration, I embarked on a scientific journey that aligned with who I am and my values.

What was the state of knowledge when you began working on these problems?

When it comes to respiratory diseases, I found that the world of public health had dogmas about the spread of pathogens in droplets of mucus and saliva from person to person. There was also this notion of pathogens spreading via aerosols — the solid residues remaining in the air after the liquid in small droplets has evaporated. But there wasn’t much modern scientific evidence regarding the behavior of droplets or aerosols. The prevailing idea was that when we exhale, the droplets that come out follow isolated trajectories — that means a pathway influenced only by gravity’s pull and the drag of air, and not by the turbulent cloud of gas that’s emitted with them. These droplets can carry pathogens (the SARS-CoV-2 virus, for example, is about 100 nanometers across, orders of magnitude smaller than the droplets).

Our work showed that a central assumption in many infection control regimes — that droplets follow isolated trajectories unaffected by the gas cloud — was wrong.

If the droplets are considered in isolation, how far one droplet will go depends only on its initial momentum. When you do the aerodynamic calculation, you get a distance of 1 to 2 meters for the larger droplets. Using the same calculations, one can show that droplets and aerosols less than 50 micrometers in size would not travel more than a few centimeters, even if the droplets are ejected at very high speeds, because of the huge drag on them relative to their size. Our work showed that the prevailing notion — that droplets are emitted in isolation and follow individual trajectories — was wrong, and that this physical picture would have to be revised to accurately assess the risk posed by droplets laden with infectious pathogens.

How did you begin studying what happens when we exhale?

At MIT, I had access to a center for high-speed imaging, the Edgerton Center, where I could use advanced imaging approaches to reveal what we can’t see with the naked eye. There are many ways to reveal what people exhale. For example, to study the liquid droplets in isolation, there is shadowgraphy or scattering, which involves imaging the shadows cast by the droplets.

I also investigated and developed other approaches to reveal not just the liquid phase but also the gas phase, including camera sensors that use a very high frame-rate to capture these extremely fast processes. The gas emitted in a breath encapsulates, traps and transports the liquid droplets, and so is clearly critical. We’re talking about emissions — a high-momentum, turbulent movement of the droplets-laden puff of warm and moist air that we exhale when we breathe, talk, sing, cough or sneeze — that occur within 100 to 200 milliseconds.

Imaging techniques can capture the light scattered by the microdroplets of the exhaled cloud, others can capture changes in air density (due to change in temperature and moisture). Combining approaches and algorithms, we can separate the largest liquid droplets from the gas puff and its cargo of droplets, some invisible to the naked eye.

These techniques allowed me to start modeling the physical process: the emergence of the exhalation and its spread in the form of a multiphase, turbulent gas cloud rather than in isolated droplet trajectories. The cloud actually governs the distance that droplets of most sizes can reach. The exhalation’s movement is initially influenced by the momentum of the gas phase and then by the background indoor airflow in a more passive, turbulent dispersal pattern.

Graphic compares the distance small and large droplets exhaled by a person might travel if calculated alone (not far) versus traveling within a turbulent gas cloud. The cloud boosts the distance for all droplets and for the smallest droplets increases the distance by a factor of 200.

Scientists interested in transmission of respiratory diseases long believed droplets from an exhaled breath would not travel far, based on looking at the droplets’ path in isolation. Newer methods have modeled the flow of the droplets within a cloud of gas and suggest smaller droplets can spread much farther than once thought.

How did you get humans to produce the necessary exhalations for the studies?

Coughing, talking, breathing is, of course, straightforward. For sneezing, it varies. Some individuals are sensitive to light and respond to it by sneezing. Others need to tickle their nose. Those involved found their own trick. Because it’s a reflex, once the sneeze is triggered it proceeds with little difference from a “natural” one.

What did you find once you did all the imaging and modeling?

I found that these earlier studies had not accounted for the presence of a gas cloud. From the point of view of fluid dynamics, most of the momentum is not in the liquid phase (the droplets). It’s mostly in the gas phase, which traps droplets within it and carries them forward in a concentrated localized packet. (That’s in contrast to the previous understanding, which was that the droplets would be spread out fairly uniformly in an indoor space.) And therefore, the overall evolution of this ejecta — its motion in space and time — depends on the physics of the gas cloud, at least at this first phase of exhalation.

Some of these drops, of course, can escape from the gas cloud and settle on surfaces, but where they escape, how they escape and where they end up, is primarily driven, again, by the physics of the cloud. The distribution, distances and timescales associated with a gas cloud laden with droplets are dramatically different than those of isolated, individual drops. The old paradigm did not account for this.

The early stage of the cloud is dominated by the very high momentum of the exhalation cloud itself, not by the background indoor airflow, which may be just a few centimeters per second and much slower than the average speed of a breath. So, initially the dynamics of the cloud dominates the dispersal of pathogen-laden droplets. The cloud can span a room in seconds to minutes. As it moves forward, the cloud draws in ambient air and slows down.

Eventually there’s a point of transition, where the exhaled cloud speed becomes comparable to that of the background air. And only then does the background airflow take over in what is a more chaotic dispersal of the droplets or aerosols that were concentrated in the respiratory cloud up to that point. This is when the concentrated packets of particulates begin to break apart and start following the pattern of airflow.

Our first observations were surprising, as clearly the reality looked very different from the existing descriptions that essentially ignored the physics of the exhaled cloud.

What kind of distances did you measure?

In the earlier scenario, small, isolated droplets, even emitted at the highest exhalation speeds, can be shown to go only a few centimeters before air resistance brings them down. Isolated, larger droplets at similar speeds are less sensitive to drag and can go further, up to 1 to 2 meters (about 3.3 to 6.6 feet).

Violent exhalations ­– coughing, sneezing, shouting or singing — may send some disease-laden droplets hurtling as much as 25 feet from the source.

How far the droplets in the cloud travel, however, is governed by the cloud’s dynamics — except for huge blobs more than a millimeter in diameter. These large blobs immediately leave the cloud. But the range of most of the smaller drops is enhanced by the gas cloud. For the most violent exhalations caused by coughing, sneezing, shouting or even singing, drops smaller than 20 or 30 micrometers across can go 200 times farther than they would if they were emitted in isolation.

In fact, the cloud and its payload can reach distances of up to 6 to 8 meters (about 19 to 26 feet) for the most violent exhalations! Even with normal breathing, the gas cloud can easily spread 2 meters with its payload of small, suspended droplets.

What happens to the gas cloud over time?

As the cloud moves forward, it sweeps up ambient air, expands and slows down. So drops moving faster than the mean speed of the cloud can escape, leaving fewer and fewer drops trapped in the cloud. When the background air flow takes over — when it’s moving faster than the cloud — the opposing forces become the ambient air flow speed versus the settling speed of the suspended particles. Droplets invisible to the naked eye, less than 10 micrometers in size (but still more than 100 times larger than most viruses), can remain suspended in the air for hours to days, depending on the background airflow.

Is this what one means by pathogen-carrying droplets being airborne?

Once you’re talking about a respiratory disease, you’re always exhaling pathogen-carrying droplets into the air. But, to cause infection, they need to be inhaled and reach their target tissue in the respiratory system. The question is one of route and level of exposure. That depends on understanding the dynamics of the gas cloud and the fate of its payload of drops and their contents, as well as how the pathogen interacts with the environment. This is dynamic, not static, so we need to incorporate such dynamic thinking about these questions to develop more robust fundamentals that can lead to improved surveillance and mitigation.

For example, the fact that SARS-CoV-2–containing droplets can remain in the air for hours with the virus potentially still being viable and dispersed indoors means that healthcare workers caring for Covid-19 patients should use high-grade respirators, and they should be putting them on well before coming face-to-face with the infected individual, not just when they are within 6 feet of the patient.

Should the rest of us wear masks?

Even at this stage of the pandemic, and given the new variants, it is key to still wear masks as an effective means of disease control in addition to personal protection. But it is important to understand that fluids follow the path of least resistance — “fluids are lazy” as we say. If a mask is not sealed — it’s open on the sides — most of the fluid passes through the largest openings, not the mask’s filter material.

However, encountering an obstacle does lower the exhalation cloud’s momentum, which reduces its range and means the cloud can be overtaken by the room’s air flow earlier in its trajectory. If most of the flow passes through the mask’s filter, as happens in well-sealed masks, what comes out is a gas flow with lowered viral particle content.  

How about ventilation in indoor spaces? What effect does it have on the spread of the droplets?

Most buildings in the US have mechanical mixing ventilation. That means that the inlet and outlet are both near the ceiling. We already know that displacement ventilation might be better at ensuring that the contaminants stay in the upper room levels rather than in the breathing zone. In displacement ventilation, cooler clean air is slowly injected from the floor or lower levels and exits from the ceiling or upper room levels. At a steady state in an ideal setting, you can create a kind of stratification, such that the breathing zone is fresher, with fewer contaminants, than the upper layer of air even with people in the room.

Obviously, in an emergency response setting, one has to work with whatever ventilation system is in place. So it is important to ensure that there’s enough fresh air coming in from the outside per unit time per person. We know from studies of tuberculosis that at least 10 to 15 liters of fresh air per second per person is needed to reduce airborne transmission of respiratory diseases indoors. That’s achievable with modern ventilation systems and even with good portable air purifiers.

Is this a concern only for hospitals or also for other indoor spaces, like grocery stores, restaurants and schools?

It’s a concern everywhere, particularly in smaller, older buildings that are not up to basic ventilation standards and that are planning to return to full or even half occupancy. Generally, building ventilation standards are not optimized for reduction of respiratory diseases, but for comfort levels. For normal occupancy during a pandemic, you need to exceed those basic standards.

You have also studied how non-respiratory infections can spread in hospitals. Tell us about that.

We looked at mechanisms that could spread spores of Clostridium difficile, a bacterium that causes serious, sometimes life-threatening infections of the colon. Hospitals are often important contributors to the transmission of this gastrointestinal disease. In North America, many hospitals use high-pressure flushes in toilets, for energy efficiency. And, again surprisingly, little work had been done on the problem of emissions from these flushes from the fluid dynamics and design points of view.

We wanted to see if the design of the devices could in fact play a role in the airborne route of transmission of C. difficile, rather than just in surface contamination. We used light-scattering and high-speed imaging and other methods to study the fragmentation — the generation of airborne droplets — from a range of flush systems.

We found a very interesting pattern. We quantified when and how contaminating droplets are created by the fluid fragmentation that is enhanced by the current designs. Plumes of small droplets are created throughout the flush process and these droplets are carried around by the background airflow and can remain suspended in a room for a long time.

Graphic shows the dispersal of water droplets from a high-efficiency toilet 5 seconds after a flush. Droplets may reach as far as 1.4 meters from the toilet bowl, with most landing in front of the toilet, and some as high as 0.8 meters above the floor. Total volume of water ejected is 6.8 milliliters.

A toilet flush can shower droplets of water around a room.

The issue we revealed is that typical cleaning protocols of hospitals may enhance such emissions. Cleaning agents, or surfactants, reduce the surface tension of the water and subsequent flushes can thus end up aerosolizing the fluid more extensively.  While certain detergents may kill viruses and bacteria, they typically do not neutralize bacterial spores. So, current toilet designs and cleaning protocols can enhance emission of such spores. 

Theoretically, those spores may end up infecting someone else. We need to be more systematic in studying these effects and developing fundamental science insights that can one day lead to improved patient-management and infection-control protocols at the frontline.

Tell us about your work with plants.

The question of contamination and disease transmission holds for animals and plants as well. I got particularly interested in the transmission of leaf diseases like rust in plants such as wheat. The connection between droplets and transmission became clear when we learned about the empirical evidence linking rainfall to the appearance a few weeks later of lesions on wheat or other primary crops.

In the lab, we started studying details of how drops of water behave as they fall on plants. The high-speed imaging revealed a rich set of processes of breakup and fragmentation of drops that had never been reported and were surprising. Most plant leaves have been thought to be super-hydrophobic, as in water-resistant, built to let water drops slide off like a raincoat. Lotus leaves are a good example of that. Yet, we found that most common crop leaves are somewhere between the extremes: not fully wetting (coated by a thin liquid film) and not fully hydrophobic. So, fluids interact with these leaves and fragment in more complex ways than would be anticipated if leaves were super-hydrophobic.

Also, leaves and stems are compliant: they move and oscillate when hit by a rain or irrigation drop. We discovered that the interaction between drops of water impacting the leaves and the wetting and mechanical properties of the leaves can cause water to fragment in a way that may be particularly effective for spreading pathogens. Disturbed by an impact, a contaminated, standing drop of water on a leaf may stretch out in a crescent shape that helps disperse any disease agents within it.

Depending on the balance of the wetting and mechanical properties, in particular the compliance or stiffness of their leaves, plants can favor short-range transmission of large drops that contain a lot of pathogens or long-range transmission of smaller droplets each containing comparatively fewer pathogens but dispersed over a greater area.

Without even knowing about the genetic susceptibility of plants to a particular or emerging leaf pathogen, one can leverage information about the dynamics of the leaves to select for crop combinations in fields. The goal is to set up contamination barriers while reducing losses in yields by designing a polyculture that integrates firewalls — crops strategically placed that are associated with a shorter range of contaminant dispersal via droplet fragmentation. 


These results involve some pretty diverse phenomena, whether it’s the transmission of respiratory diseases, or the spread of infections in hospitals, or transmission of diseases in plants. Is there a common theme?

All of these insights are linked by their fascinating fluid dynamics and interfacial physics — what happens when fluids and solids meet. It’s curiosity-driven and focused on fundamentals. When a crisis hits, it is not obvious early on what kinds of basic research will become important. So, it’s crucial to support research that may not appear ready for immediate use.

The type of research I do, and did for years prior to this pandemic, is focused on the intersection of fundamental fluid dynamics, biophysics and infectious disease. It was not particularly popular, mainstream nor funded by traditional sources. Nevertheless, we carried on, and the insights we gained turned out to be central to key safety measures and led to an explosion of research in this area that will enable us to better prepare and respond to future crises.

When we face new challenges, it is often the insights from basic, scientist-driven research that can enable or suggest solutions. That is why it is so vital to be wary of group-think and nurture intellectual freedom and diversity in the research enterprise.

Given that you are studying the transmission of respiratory diseases, how have the many months of the pandemic been for you?

Very busy and grueling. There’s still a sense of urgency, particularly given the resurgence of Covid-19 infections with the fourth wave and with the newer, highly transmissible variants. But there is also a sense of moral obligation and duty to educate, communicate and share in any way we can. This is a mission way beyond the usual ivory tower of academia. The pandemic and the knowledge needed to combat it are both still unfolding. Staying focused on giving back to society should be core to the mission of universities, particularly in this time of need.

Editor’s note: This article was revised on September 6, 2021, to correct two errors. The piece should have said that the previous understanding was that droplets, not the gas cloud, would be distributed fairly uniformly through a room. In addition, the description of a droplet's trajectory considered in isolation should have said that it would not have been influenced by the gas cloud emitted with it as opposed to being influenced by other droplets, as was stated originally.

This article is part of Reset: The Science of Crisis & Recovery, an ongoing Knowable Magazine series exploring how the world is navigating the coronavirus pandemic, its consequences and the way forward. Reset is supported by a grant from the Alfred P. Sloan Foundation.